FIG. 1 is a hardware structure diagram of a storage system in the prior art. As shown in FIG. 1, in this storage system, a user host 1 and a disk chassis 2 are each directly connected to its adapter card 3; and the adapter card 3 is connected to a north bridge 4 through a PCIe bus (one of extended personal computer bus interface specification PCI buses), and is further connected to a central processing unit (CPU) 5. Because the CPUS does not have a direct bus interface, the CPUS may be connected to peripherals (such as the user host and the disk chassis) through a chip of the north bridge 4. A function of the adapter card 3 is to implement a specific physical link protocol, so that a request is sent or received through a physical link. Besides, a memory chip 6 is configured for the CPUS exclusively. The adapter card connected to the user host 1 is called a front-end adapter card; and the adapter card connected to the disk chassis 2 is called a back-end adapter card.
The foregoing north bridge 4 and the CPU 5 form a control board. In the prior art, the control board serves a group of user hosts located at a front end. Meanwhile, several disk chassis are configured, so that an independent storage subsystem including a control board is formed. That is, in a storage subsystem, a user host can request only one control board to store or read data, and a disk chassis is controlled by only one control board.
In the existing storage system, if a problem occurs on the north bridge or the CPU of its control board or the connection, for example, when the CPU is damaged or in hibernation, it causes the front-end user host and the disk chassis that exist in the same storage subsystem as the CPU to fail.